TWI343522B - Non-volatile memory storage systems and methods for performing phased garbage collection - Google Patents

Abstract

A method for operating a non-volatile memory storage system is provided. In this method, a write command is received to write data. The write command is allocated a timeout period to complete an execution of the write command. Within the timeout period, a portion of a garbage collection operation is performed. The data associated with the write command are written to a buffer associated with the non-volatile memory storage system.

Description

1343522 IX. Description of the Invention: TECHNICAL FIELD OF THE INVENTION The present invention relates generally to memory operations and, moreover, to methods and systems for performing a phased waste project collection operation. [Prior Art] In a non-volatile memory storage system, a data block stored in a memory is collected (i.e., compressed or combined) by a periodically discarded item to recover the storage capacity of the memory. In a typical abandoned project collection operation, valid data is copied from the block to another block. The original block is erased in the transfer of valid data to provide storage capacity. Currently, the write operation can be triggered - the non-volatile memory storage system is executed - the waste collection operation. The host allows a fixed amount of time to perform a writer operation including the collection of the discarded item. For example, the secure digital agreement limits the amount of time to 250 milliseconds. If the non-volatile memory storage system exceeds this fixed amount of time in a write operation, the size of a memory block that can be generated has increased due to increased capacity, linearity, and grain size. Therefore, because more data is transferred, it takes longer to perform a write operation. - The waste symplectic can easily exceed the fixed amount of time that is configured to be used for write operations and therefore used directly for the intrusion operation. Therefore, it is necessary to prevent a time-out error when the amount of time spent performing a waste bin for a predetermined amount of time exceeds the solid content. [Invention] Various embodiments of the present invention provide a method for collecting waste items in stages and / or system. It will be appreciated that the embodiments can be implemented in numerous ways, and that the I2273I.doc 1343522 is embodied in a number of ways, as a method, an electrical system, a system, or a device. Several embodiments of the invention are described below. In accordance with an embodiment of the present invention, a method for operating a non-volatile memory storage system is provided. In this method, a write command is received to write data. The write command is configured with a cold n set and a timeout period to complete execution of the write command. Performing an abandoned project collection operation during the timeout period

:, -section. The data associated with the write command is written to a buffer associated with the non-volatile memory storage system. Other embodiments and advantages of the present invention will be apparent from the following description of the accompanying drawings in conjunction with the accompanying drawings. FIG. The detailed description of the one or more embodiments is provided in conjunction with the accompanying drawings, and the detailed description of the embodiments of the invention, and the detailed description is not limited to any specific implementations. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the details of the invention. For the sake of clarity, the technical material associated with the embodiments that is known in the art is not described in detail to avoid unnecessarily obscuring the description. The embodiments described herein provide methods for phased collection of waste projects and/or Or system. In general, an abandoned project operation can be split into multiple phases. It can be over multiple timeout periods. Execution of the waste project collection operation 122731.doc -6 - : paragraph (or part). In the embodiment, as will be explained in more detail below, in a time-out period - part of the waste collection operation can be The data received from the write command is stored in a buffer. The figure is a simplified block diagram of a non-volatile memory storage system according to an embodiment of the present invention. A host system (for example, a desktop) Electricity: 9 Λ player, digital camera 'and other computing devices' can write data to non-volatile memory storage system 102 and read data from non-volatile memory storage system i 0 2 . Non-volatile memory The storage system i 〇 2 can be embedded in the host or removably connected to the host. # Shown in Figure 1, the non-volatile memory storage system 1〇2 includes memory that communicates with the memory ιι8# & 〇°-likely, the 'memory controller i 10 controls the operation of the memory U8. Examples of operations include writing (or stylizing) data, reading data, erasing data, verifying data, and performing waste collection operations. , and other #作.记The memory controller 110 includes a bus bar 224 that establishes an interface with the system bus 126 via the host interface 1-4, and the memory controller establishes an interface with the memory 丨8 via the memory interface 丨08. The host interface 1 〇4, processor 106 (eg, test processor, microcontroller, and other processors), memory interface 108, random access memory (RAM) 112, error correction code (ecc) circuit 114, and read only The memory (ROM) 1 1 6 communicates by means of the bus i 24 .

The ROM 116 can store a storage system firmware including program instructions for controlling the operation of the memory U8. Processor 106 is configured to execute program instructions loaded from R〇M 116. The storage system firmware can be temporarily loaded into the ram 1 2, and in addition, RAM can be used to buffer data transferred between a host and the memory 1 18 . The ECC circuit 1 14 can check for errors passed between the host and the memory 118 via the memory controller 122731.doc 1343522 no. If an error is found, the ECC circuit 114 can correct a number of error bits, depending on the caching algorithm utilized. The memory 11 8 may include an array logic 12 〇, a non-volatile memory cell array 122, and an A memory cell array 123. Non-volatile memory Zhao cell array 】 22 can include a variety of non-volatile memory Zhao structure and technology. Examples of non-volatile memory technologies include flash memory (eg, NAND, n〇r, multi-level single-turn (MLC), divided bit lines N〇R (DIN〇R), AND, high capacitance coupling ratio ( HiCR), asymmetric contactless transistor (ACT), other flash memory), erasable readable read-only memory (EpR〇M), electronic erasable programmable read-only memory (EEPR〇) M), read-only memory (r〇m), one-time programmable memory (〇τρ), and other memory technologies. In one embodiment, memory 118, in turn, can include a memory cell array 123 configured to store a buffer. The buffer is configured to store data in a fractional slave project collection operation. The new data received from a write command can be stored in the buffer in a staged discarding project collection operation as will be described in more detail below. It will be appreciated that the buffers may also be located in RAM 112 and/or non-volatile memory unit P trains 122 in accordance with embodiments of the present invention. Similar to the non-volatile memory cell array 122, the memory arrays 123 can include a variety of memory structures and techniques. ◎ Since the memory single array 123 is configured for buffering operations, the memory cell array can be incorporated with a comparable The non-volatile memory array 122 is faster, more economical, and more versatile than the different memory structures. The array logic 120 establishes an interface with the memory controller u and the non-volatile memory unit 122731.doc 1343522 and the memory cell array 123, and can provide, for example, a non-volatile memory cell array and memory. Addressing, data transfer and sensing of cell arrays, and other support. To support the non-volatile memory cell array m and the memory cell array 123, the array logic 12A may include a column decoder, a row decoder, a charge pump, a word line voltage generator, a page buffer, an input/output buffer, Address buffers, and other circuits.

2 is a simplified block diagram of a memory cell array organized into planes. The one or more memory cell arrays can be divided into a plurality of planes or sub-arrays. In the example of Figure 2, a memory cell array is divided into four planes 2〇2_

. It will be appreciated that there may be other numbers of planes (e.g., i, 2, 8, 6 or more) in a non-volatile memory storage system. Each plane 202, 203, 204 or 205 may be divided into memory unit blocks, such as blocks 210-213 and 220-223 located in respective planes 202-205. A memory unit block is the minimum number of memory units that can be physically erased together. For increased parallelism, operations may be performed in a larger metablock unit in which blocks from each plane 202, 203, 204, or 205 are logically linked together to form a unitary block. These blocks. For example, four blocks 210-213 can be logically linked together to form a unitary block. Further, the blocks used to form the unary blocks may come from various locations within their respective planes, such as planes 202-205. For example, four blocks 220-223 from various locations within respective planes 202-205 can be logically linked together to form another metablock. The unary block may extend over all four logical planes 202-205 within the non-volatile memory storage system, or the non-volatile memory storage system may be one or more of one or more different planes 122731.doc - 9- 1343522 blocks dynamically form metablocks. Figure 3 is a simplified block diagram of a memory cell page. Each block (such as block 210-2 13) is further divided into memory unit pages. As shown in Figure 3, each block 210, 211, 212 or 213 is divided into eight pages p 〇 P7. Alternatively, there may be 16, 32 or more memory cell pages within each block 210, 211, 212 or 213. To increase the operational parallelism of a non-volatile memory storage system, pages within two or more blocks can be logically linked into metapages. For example, a unitary page may be formed from a page (e.g., ρι) from each of the four blocks 210-213. The unary pages may extend over all planes within the non-volatile memory storage system, or the non-volatile memory storage system may be from one or more of the one or more separate blocks The meta page is dynamically formed. 4 is a simplified block diagram of a section of a memory cell. A page can be further divided into one or more segments. The amount of data in each page can be an integer multiple of one or more data segments, and each of the #_ segments can store 5 12 bytes of data. Figure 4 shows page 4〇1 divided into two sections 4〇2 and 4〇4. Each segment 402 or 404 contains data 4 〇 6 that can be 512 bytes in size, and additional information 4 〇 5 associated with the data. The additional data 4〇5 may be a 16-bit group and may store, for example, one of the ECCs calculated from the data 4〇6 during the stylization, the logical address associated with the data, the block has been erased, and The number of leaves, control flags, operating voltage levels, and other information associated with the data. Figure 5 is a simplified block diagram of a logical "face between a host and a non-volatile memory storage system. - The continuous logical address $ is provided for the bit b of the data stored in the 6 memory. If the logical address is viewed by the host, the workstation 512 can be divided into incremental data clusters. Each __cluster can include a number of data segments, such as between 4 segments and 64 segments. As shown in Figure 5, the application executing on the host creates three data rights 1, 2 and 3. Cases 2 and 3 can be a sequenced collection of data and are identified by a unique name or other reference. The host assigns a logical address space to the file i, which has not been located to other files. Here, the file 1 has been assigned a continuous range of available logical addresses. When the field host creates file 2 after file 1, the host similarly assigns two different ranges of consecutive addresses within logical address space 512. The host may not assign a contiguous logical address to a file (such as file 1, 2 or 3), but may assign a slice 1 of a logical address between logical address ranges already configured for other files. The example of Figure 5 shows a non-contiguous address range of another file 3 configured logical address space 5^2 that was not previously configured for the file and 2 and other data. The host can remember the logical address space 512 by maintaining a file configuration table (FAT), which maintains the logical address assigned to the various data files, such as files, by the host by conversion. The host refers to the file U based on the logical address and not based on the physical location of the non-volatile memory storage system. On the other hand, the non-volatile memory storage system refers to files 1-3 based on portions of the logical addresses to which the data has been written and does not refer to the files based on the logical addresses assigned to the files. The non-volatile memory storage system converts the logical address provided by the host into a unique physical address within the memory cell array 502 that stores data from the host. Block 5〇4 represents the table of these logical to physical bits 122731.doc • Π - 1343522, which is maintained by the non-volatile memory storage system. 6 is a flow diagram of a general overview of operations for phased-dump project collection in accordance with an embodiment of the present invention. It should be understood that when the original stored data is invalid, the new data can be overwritten with the information stored at the specific host logical address. In response to this, the non-volatile memory storage system writes new data into an update block and then changes the logical address to the physical address table for their logical addresses to identify new physical areas through which new data is stored. Piece. The blocks containing the original data at their logical addresses are then erased and made available for additional storage. This erasure can occur prior to a write operation. Therefore, the new data is written to the same logical address until the memory controller learns that the data at a given logical address has been invalidated or invalidated by the host. Many memory blocks can therefore store invalid data for a period of time. The size of the block and the metablock is increasing and such an increase causes most of the individual data to be written into the data storing the storage capacity smaller than the one-dimensional block, and in the case of Xu Xi, the storage is even less than one block. The amount of capacity is the φ data. Because the non-volatile memory storage system can direct new data to an erased metablock, the boot can cause portions of the block or metablock to be unfilled. If the new information is for some of the assets stored in another block, the #update's will be from the remaining blocks of the data from the other blocks with the logical address adjacent to the logical address of the new data element. The valid metapages are also copied to the new metablock in logical address order. The old metablock can retain 7L pages of other valid data. Therefore, the data of a particular metapage of a unique block can be invalidated or replaced with new data with the same logical address written to a different metablock. 122731.doc -12- 1343522 Maintaining sufficient physical memory space for storing data on a logical address space 'can be periodically collected (ie, compressed or merged) for such data. In general, an abandoned project collection operation involves The self-block reads the valid data and writes the valid data to a new block, and the invalid data in the process of the heart. For example, in the block diagram of Figure 5, the new data file 3 is created to make the old data file 3 Cancel the old data file 3 to recover the physical capacity used by the old data file 3. However, if the case

2 and the old file 3 is stored in the same physical block, then the erase operation will trigger - the waste collection operation. Returning to Figure 6, the non-volatile memory storage system can perform an obsolete item collection operation within a timeout period configured for a write command. According to an embodiment of the present invention, if the waste collection operation cannot be completed within a timeout period, an abandoned project collection operation can be split into several different stages (or portions). Here, the non-volatile memory storage system performs a portion of the waste collection operation using a timeout period configured for a plurality of write instructions.

In other words, the non-volatile memory storage system performs a portion of the waste collection operation using a timeout period configured for a plurality of write instructions. As shown in Figure 6, a write command for writing a new asset is received in operation 602. As used herein, the term "new data" is defined as data that is received by a non-volatile memory storage system from a write command to write to a file. A one-time period is configured for the write command to complete execution of the write command. In other words, the timeout period is a time period configured for the execution of the write instruction. An example of a configured timeout period is 250 seconds. The write command can be a single segment write instruction or a multi-session write instruction. 122731 .d〇c -13- 1343522 As will be explained in more detail below, in a single-segment write command, new data can be written as a single segment to a random address on a memory. In a multi-session write instruction, new data of multiple sectors with adjacent logical addresses is written to the memory. The right discarded project collection operation may not be completed within the timeout period, and as shown in operation 604, a portion of the discarded project collection operation is performed within a timeout period configured to write the instruction. The remainder of the collection of abandoned items can be completed in the post-week period. By way of example, Figure 7 shows a simplified block diagram of an example of a waste collection operation that is split into multiple portions 78A and 78i, in accordance with an embodiment of the present invention. As shown in Figure 7, a non-volatile memory storage system receives the multi-session write command 704' and then receives new data 76G-762 for storage into a plurality of segments in the memory. The busy recording signal 7G2 is maintained after receiving each data section·, 761 or addition. The non-volatile memory material n maintains the busy recording signal 702 to allow execution of the write command 'which may include a discarding item collection operation (if needed) and other operations. When the busy recording signal is maintained, a host will not send another command or additional data to the non-volatile memory system. The non-volatile memory storage system can maintain the busy recording signal for a limited amount of time after receiving each of the four (four) segments 760, 761 or 762, since the host allows a limited amount of fixed time (ie, the 'timeout period MO' It is used to execute the write command. ^The signal in the busy recording is longer than the timeout period of 75〇, Sichuan or 752, then the host can repeat the write command or give up the process. Therefore, the non-volatile memory storage system cannot be The busy recording signal is added during the period longer than the timeout period (four), 751 or 752. After the completion of writing more 122731.doc 14 丄州!) 22 data sections 760-762 release the busy recording signal 7〇2 The host is allowed to further communicate with the non-volatile memory storage system. # Referring to Figure 7', portions 780 and 781 of the discarded project collection can be configured between multiple timeout periods of 75G.752. In other words, the non-volatile memory storage system utilizes a per-time period of 750, 751 or 752 execution------------------------------------------ For example, f, during the first-timeout period 750, the first-part 780 of the collection of obsolete items is performed. Here, during the φ "timeout period 750, one of the valid data portions may be copied from one block to another. In the second timeout period 751, the previous start from the first time period is continued. The waste collection operation. During the timeout period of 75 ,, the non-volatile memory storage system performs the second part 781 of the previous abandoned project collection operation until the previous collection of the abandoned project is completed. The remaining part or the last part is copied from one block to another to complete the previous collection of abandoned items. If the previous abandoned item collection operation cannot be completed within the second timeout period 751, the non-volatile memory is stored. The system may complete the obsolete item collection operation using a subsequent timeout period (such as a third timeout period 752). At the end of the multi-session write instruction 704, the non-volatile memory storage system may receive a stop instruction After 7〇6, the busy recording signal 702 is maintained until all data sectors 76〇762 are written to the memory cell array. Please note that FIG. 7 illustrates the multi-segment write instruction. Operation. As will be explained in more detail below, for single-region write slashes and multi-session write commands, the discarded project collection operations performed are different. For example, as will be explained in more detail below, The type of buffer for storing new data is a single-segment write according to the received write command. 122731.doc •15· 1343522 曰7 is also dependent on the multi-segment write command. Return to Figure 6 'in one Execution of waste in the overdue period. (4) The data collection operation of the obsolete project during the period is stored in the operation_, and the new '=; Xiao non-volatile 5 memory storage system can be received from the write operation. Associated Buffers - In an embodiment, the buffer may be a data structure associated with a single column of non-volatile memory (e.g., non-volatile memory array I shown in Figure). Examples of data broadcasts, mussels, and frames are described in more detail below.

The blocks of the non-volatile memory cell array (such as write buffer blocks) are described. In other embodiments, the buffers can be blocks of a volatile memory cell array. For example, 'new data can be stored in a block in a RAM associated with a non-volatile memory storage system (e.g., RAM 112 shown in Figure 1). In the re-embodiment, as described above, the new material can be stored in a block located in an array of individual memory cells (e.g., memory cell array 123 shown in Figure i).

Figure 8 is a flow diagram of detailed operations for performing a staged collection of obsolete items in accordance with an embodiment of the present invention. As shown in Figure 8, a write command is received in operation 802 to write new data to a memory. In one embodiment, the write command is a single segment write command. In some cases, as will be explained in more detail below, according to another embodiment, the write command can also be a multi-session write command. After receiving the write command, in operation 804, the non-volatile memory storage system maintains a busy recording signal. Prior to executing the write command, in operation 806, a portion of the obsolete item collection operation is performed in an obsolete item collection time period. For example, in one embodiment, one or more blocks are selected for use in the collection of obsolete items 122731.doc -16-1343522. The one or more blocks may include invalid data and/or valid data. The valid data is copied to a second block during the collection period of the abandoned project. Copy all valid material or part of valid data to the second block based on the collection time period of the abandoned project collection and the amount of valid data to be copied for the waste project collection operation. The collection time period of the abandoned project configured for the collection of waste projects can be expressed as the collection time period of the abandoned project = timeout period _Tpr〇g

Its _ timeout period is H as limited time (four) as described above.

Tprog is the maximum programmed time associated with writing new data to memory (for example, to a 'buffer'), or the maximum cost of writing new data to memory for a non-volatile memory storage system. time. Because &, in one embodiment, the non-volatile memory storage system follows the amount of time during which the valid data is copied from the one or more blocks to the second block. The non-volatile memory storage system stops replication until the time exceeds the collection time period of the abandoned project. If the obsolete item collection operation cannot be completed during the obsolete item collection time period, then new data associated with the writer instruction can be written to the write buffer block in operation 81. Non-volatile memory storage systems are available for disposal. Write new data to the write buffer block before, during, or after the collection operation. As will be explained in more detail below, once the waste project is completed, the new data can be copied from the write buffer block to the update block. The non-volatile memory storage system will write the buffer area. Keep in the body of the recall. As a general matter, the write buffer block buffers non-volatile memory 122731. (10, 1343522 new data in the body storage system. In one embodiment, the write buffer block spans multiple logical addresses. In another embodiment, the write buffer block spans an entire logical address space. By spanning the entire logical address space, the write buffer block can be stored for writing to the entire non-volatile memory storage system. All logical addresses and data of all logical address groups (ie, all logical groups). In other words, new data associated with different logical groups can be stored in the write buffer block. The group is a group of logical addresses having a size equal to the size of the unary block. New data can be written to the write buffer block in a continuous or discontinuous (i.e., unordered or random) order. As will be explained in more detail below, when a new data written to a write buffer block is later copied to another block (eg, an update block), the write buffer block acts as a temporary buffer. As used in this article The term "write buffer block clear" means copying new data stored in the write buffer block to another block. Note that in one embodiment, the non-volatile memory storage system The write buffer block is used for phased project collection operations. Therefore, the write buffer block may not store the most recently updated new data. Therefore, if a new data is read in a subsequent read operation, The non-volatile memory storage system then checks which new data stored in the update block or write buffer block has been updated recently. The read operation then accesses and returns the most recently updated new data. In another implementation In the example, the non-volatile memory storage system can use the write buffer block for the non-staged project collection operation and the staged waste collection operation. Here, the write buffer block is used for all writes. Into the operation. In other words, the new data from the non-staged project collection operation 122731.doc -18-2 staged project collection operation is written to the write buffer area: therefore 'write buffer The block operates as if it were written to the cache memory and thus includes the most recently updated new data. The index information associated with the Xiao write buffer block can be stored in a separate resource. Into the buffer block itself. The data structure can be stored in the RAM of the memory controller. Stored in the data structure ^ 纟 ## allows to track and access the write buffer block • Index information (for example) may include an index of a logical address to one of the locations in the write buffer block or a valid index of the write buffer block. After closing the update block or in the pair After updating a number of updates to the buffer block, the index information is updated. As will be explained in more detail below, in one embodiment, the write buffer block can be configured for a sector level index. In the example, the 'write buffer block' can be configured for a one-page boundary index. The write buffer block can also be compressed when the write buffer block is full. J. The valid data stored in the write buffer area is copied to a new block that is referred to as a new write buffer block, and the existing 2 write buffer block with the invalidated item is erased. As used herein, the term "write buffer compression" means compressing the payload data stored in the write buffer block. • Referring again to Figure 8' after the new material is written to the write buffer block and the waste item collection operation is performed during the waste project collection time period, the non-volatile memory storage system expires in the timeout period. prior to. Record the nickname. Therefore, the total time for executing a write command including a waste project collection operation or a waste project collection operation 122731.doc -19-1343522 does not exceed the timeout period. If part of the waste collection operation is performed within the timeout period, the remainder is completed in the subsequent timeout period. When the disposal of the waste project is completed, the collected or discarded blocks of the abandoned project are erased and made available for the storage of additional data.

9A and 9B are simplified block diagrams of memory blocks having successively updated blocks collected by staged obsolescence items, in accordance with an embodiment of the present invention. As shown in Figure 9A, the original block A_ and associated consecutive update block A 982 are selected for obsolete item collection. In general, the data received from the write command can be written to a new area. A dedicated metablock may be assigned to each logical group in which the data is updated as an update block. Note that the logical data section is stored in a logical group containing a set of logically adjacent sections. An update block can be managed to receive data in a sequential or out-of-order order (i.e., in a discontinuous order). It should be appreciated that a contiguous update block, such as contiguous update block A 982, is a metablock configured when a write command is received from the host to write data, and the data is populated in a logical group.

One or more physical pages, all valid segments of the logical group are currently in the same metablock. The data sectors written to the contiguous update block are contiguously written logically addressed such that the segments are instead written to the corresponding logical segments in the original block. One of the updates in this logical group can be written to this successive update block until the consecutive update block is closed or converted to an out-of-order update block. Note that when the last physical segment location of the successive update block is written, the continuous update block is considered closed. In other words, the closure of the contiguous update block can be caused by the contiguous update block being completely written by the host or populated with updated section data copied from the original block. As explained in more detail below, 122731.doc • 20· 1343522, when a data segment written by a host is logically discontinuous with a previously written data segment within the logical group being updated, An out-of-order update block is created by a continuous update block conversion. The original block _ may include the shadow pattern and the dot shape in Fig. 9

The pattern indicates invalid and valid data. Note that in addition to the valid material copied from the original block A_, the continuous update block A additionally includes the existing material 981 written to the continuous update block A before the discarded item collection operation. When receiving - for writing a new data such as a write command, the write command may trigger the closing of the continuous update block A 982 (which is an abandoned item collection operation type), a is a new data and one does not have a Logical group association or new f-material call of the open update block—discarded item (4) operation. The non-volatile memory storage system maintains the __busy t signal and then copies the valid data from the original block A 98〇 to the continuous update block A 982 until the first obsolete item collection time period 97〇 expires. During the replication period, the non-volatile memory storage system tracks the time, and the non-volatile essay storage system exceeds the first waste item collection time period of 97 ;; the copy operation is stopped. As shown in Fig. 9A, the waste project collection operation cannot be completed within the first-depleted project collection time period of 97 , because the valid data remains in the original block A 98〇. Therefore, after copying the valid portion to the continuous update area 982, the new time 983 is allowed to be written to the buffer block 908 before the expiration of the _th timeout period. Figure 9B shows that the remainder of the waste collection operation can be completed in one cycle. Here, 'received after the first-write instruction' is used to write the second write instruction of 12273l.doc -21 - 1343522 new material 984. _ ^ ^ Therefore, the second write command is configured with a second timeout period. During the second timeout period, the remaining valid data: original block _ is copied to the continuous update block α 982. In this example η, all remaining effect data (or the last part of the valid data) is copied to the continuous update block α 82 in the first-deprecated item collection time period 972 so that the waste collection can be completed in the second timeout period. operating. Since the continuous update block 982 is filled, the continuous update block A is converted into a new original block A state or a non-update block. Since the discard item collection operation is completed in this second timeout period, the original block A 98G is erased and made available for storage of additional data. After erasing the original block A 980, the more advanced block c (10) is configured and the new data 984 received from the second write command is written to the newly configured update block c. Please note that the update block C 986 may or may not be associated with the new original block A 985. After the collection of the obsolete items is completed in the second timeout period, if there is an available time in the second timeout period, the non-volatile memory storage system can perform a write buffer block clearing operation. When a new material (such as new material 983) written to the write buffer block is later copied to another block (for example, update block C 986), the write buffer block 908 acts as a temporary buffer. In the example of Fig. 9B, there is a time for a write buffer block clear operation in the second timeout period. The new material 983 stored in the write buffer block 〇8 is in the same logical group as the new material 984. Therefore, the new material 983 is copied to the update block C 986 after the completion of the waste project collection operation. The new material 983 stored in the write buffer block 908 is marked as invalid 'and thus allows additional space in the write buffer block to be used for the storage of additional new data 122731.doc -22 1343522" Note that a write buffer block clear operation may not be performed after each obsolete item collection operation is completed. The timing of writing to the buffer block clear is explained in more detail below. Figures 10A through 10E are simplified block diagrams of memory blocks of an out-of-order update block collected in accordance with the present invention. As shown in FIG. 1A, the original block A 902 and the out-of-order update block A 904 are selected for discarding project collection. In general, one is like disorder

The out-of-order update block of update block A 904 allows the data sectors to be updated in a random order within a logical group, and any repetition of individual segments can occur. An unordered update region can be established by a continuous update block transition when a data segment written by a host is logically discontinuous with a previously written data segment within the logical group being updated. Piece. All data segments updated in this logical group are written to the next available segment location in the out-of-order update block, regardless of the logical segment address within the data segment and within the data segment. .

Here, the original block A 902 and the out-of-order update block 八 〇 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓 枓Although the receiving one is used to write new capital, the non-volatile memory storage system maintains a busy record when the order is entered.

The original block A 902 and the I sequence update ": the receiver will validate the data until the first waste: the update block A 9〇4 is copied to the new block A f period, the set time period 950 expires. Copying the 'non-volatile memory storage memory storage system to stop the copy operation before the first waste symplectic term = 'between' and non-volatile. As shown in the figure, the time period of 95 〇 I22731.doc -23· 1343522 The existence of the hot first waste project collection time period 950 is completed, because the data is still valid after the original block A 9〇2 and the disorder update block A 904 are effective. Part of the data is copied to the new block A 906. Before the expiration of the data cycle, 'the new data received before the waste project collection operation, 9〇1, is written to the write buffer block.

Figure_ shows the lapse of the __ subsequent timeout period of the culling project collection operation: the cage receives a second write command after the first write command and configures the write command to the second timeout period. The remaining valid data is copied from the original block A 9〇2 and the out-of-order update block A_ to the new block A 906 during the second timeout period until the second discarded item collection time period 95] expires. As shown in Figure _, the abandoned project collection operation cannot be completed within the second abandoned project collection time period 951 'because there are still valid data remaining in the original block A 902 and the out-of-order update block a 9〇4 After copying a part of the valid data to the new block A 9〇6, the new data 9〇5 received before the start of the waste project collection operation is written to the write buffer block 9〇8. Figure 10C shows that the remainder of the waste project collection operation can be completed in a third timeout period. A third write command is received after the second write command. Therefore, a third timeout period is configured for the second write command. During the third timeout period, the remaining valid data is copied from the original block A 9〇2 and the out-of-order update block A 904 to the new block a 906. Here, all remaining valid data (or the last portion of the valid data) may be copied to the new block A 906 during the third scrap item collection time period 953. Therefore, the waste collection operation can be completed in the third timeout period. Because the waste collection operation is at 122731.doc -24· 1343522

This third timeout period is completed, so the original block A 902 and the out-of-order update block A 904 can be erased and made available for the storage of additional data "Erasing the original block A 902 and the out-of-order update area" After block a 904, the new update block B 962 is configured and the new data 9〇9 received from the third write command is written to the newly configured update block B. New material 909 is not written to write buffer block 908 because the obsolete item collection operation is complete and does not need to buffer new data before a subsequent timeout period arrives. In this example, a write buffer is executed. The block clear operation ' copies the new data 901 and 905 stored in the write buffer block 908 to the update block β 962, assuming that the new data 901, 905 and 909 belong to the same logical group. Figure 10D shows another discarding operation performed during a fourth timeout period. As shown in Figure 10D, the original block Β 96 〇 and the unordered new block 962 962 are selected for waste project collection. The updated block 962 962 that has been converted into a sequential update block includes new data 901, 9〇5, and 9〇9 associated with the first, second, and third incoming instructions. When receiving a fourth

When the command is entered, the non-volatile memory storage system maintains a busy recording signal and then valid data is copied from the original block 960 960 and the unordered update block 962 962 to the new block 964. As shown in Figure 10D, the discarded project collection cannot be completed within the fourth abandoned project collection time period 952 because there is data remaining in the original block 960 and the out-of-order update area 962. The new data 9〇3 received from the fourth write command is written to the write buffer block 908 after copying a portion of the valid data to the new block β 9Μ. Figure 10A shows the waste project of Figure 10D. The remainder of the collection operation can be completed in a fifth timeout period of 122731.doc -25· 1343522. A fifth write command is received after the fourth write command. Therefore, a fifth timeout period is configured for the fifth write command. During the fifth timeout period, the remaining valid data is copied from the original block B 960 and the out-of-order update block B 962 to the new block B 964. Here, all remaining valid data (or the last part of the valid data) can be copied to the new block B 964 during the fifth abandoned project collection time period 953. Therefore, the waste collection operation initially performed during the fourth timeout period can be completed in the fifth timeout period. Since the discarding item collection operation is in this fifth timeout period, the original block B 960 and the out-of-order update block b 962 are erased and made available for the storage of additional data. After erasing the original block B 96 〇 and the out-of-order update block B 962, the update block c 965 is configured and the new material 907 received from the fifth write command is written to the newly configured update block c. Please note that update block C 965 may or may not be associated with new block B 964. Here, a write buffer block clear operation is performed, and therefore, the new material 903 stored in the write buffer block 908 is copied to the update block c 965, assuming new material 903 and new data 9〇7 Belong to the same logical group. 11 is a flow diagram of an operation for optimizing access to a write buffer block in accordance with an embodiment of the present invention. Write to the buffer block (4) into some additional consumption. To reduce the additional overhead, in one embodiment, the write instructions shown in the write <write" of the write buffer block can be minimized. Thereafter, in operation _ the non-volatile body storage system estimates - the execution phase 1 execution time is the time period for the write command. The execution of the write command can program (or write) the new data to the memory "execution - the collection of discarded items", and perform other operations. For example, the execution time may include a time period for copying valid data from one or more blocks into another block and a time period for writing new data. The non-volatile memory storage system estimates the execution time in real time. The non-volatile memory storage system can estimate execution time based on various parameters. Examples of parameters include obsolete item collection types (eg, block closure, consolidation, compression, and other types), valid data stored in blocks to be collected by obsolete items (eg, update blocks and other blocks) Quantity, the stylized time associated with the non-volatile Lum memory system, the size of the obsolete item t-size (10) such as 'the size of the meta-block', the pre-filled, pre-populated data Quantity, pipeline, cache, frequency staging failure, and other parameters. In the estimation, one or more of the first blocks may include one of the eleven pages of the time, and the instance data will be copied to a second block. - The amount of time spent on the non-volatile w/ volatile memory storage system ^ / body storage line can be read by the amount of non-volatile bamboo and the first bn The storage system adds the amount of valid data written in time to estimate the execution time. After the estimated execution time of the non-volatile memory (four) memory system, the execution time and the over-operation are shown in 1 10〇6, if the estimated execution time is disputed. As shown in operation 1008, the main contract is less than the timeout period or the fall can be completed within the discarded project = abandonment, the collection operation and the timeout period - update the block. Call) and write new data to .X.. As shown in operation 1010, if the execution time of the # is more than the timeout β right estimate period (ie, greater than or equal to the timeout week -27 .)) then write new data to the write buffer block, and can use the configured obsolete item collection time period for writing buffer block clearing or for performing part of the waste project collection operation (if In an embodiment, as will be explained in more detail below, the write buffer block clear may begin with the first valid entry of the write buffer block for the remainder of the timeout period. By writing to the write buffer block when the estimated execution time exceeds the timeout period, there will be less writes to the write buffer block and, therefore, associated with the hold write buffer block The additional consumption can be reduced. Into the buffer block - page Boundary Index In one embodiment, a write buffer block is constructed to include multiple pages and each page within the write buffer block is indexed. Each page of the write buffer block is configured to Store new data for one or more segments. Therefore, an indicator points to or refers to a page' and can refer to multiple segments in a page by an indicator because the indicator points to a page and does not point to the page. A plurality of sections. Here, new data belonging to the section of the same metapage is written to the same metapage of the write buffer block. However, new data of the sections belonging to different metapages are written. Into the different metapages of the write buffer block. For example, t, one of the first: the new data of the section is written to the new buffer of the write buffer 。·. Incoming command receiving - second area: capital. If the new data of the second section belongs to the same meta page, the second section is written with a new page. An ancient product. To the write buffer The first part of the block will be the first if the new data of the second sector belongs to the different meta page, and the second sector is written to the write buffer block - A common page. 122731.doc • 28· Write Buffer Block - Segment Level Index In another embodiment, the write buffer block is constructed to include multiple regions U written into the buffer block. One or more segments may be indexed. For example, an indicator may point to or reference multiple segments (eg, four segments). In another example, the indicator may point to or reference a segment. In the case of a per-segment indexed in the write-and-buffer block, the data is written to a single segment of the write buffer block. Subsequently associated with the other_write instruction New data is written to the next available segment in the write buffer block. For example, new data from segments of multiple write instructions can be written to the same in the write buffer block Different sections of the page. Thus the write buffer block can be constructed to include multiple sections and each section is configured to store new data for a section. In order to maintain single-segment writes without a time-out error, the non-volatile memory storage system can configure a number of segments to a write buffer block such that the configured segments are larger than the non-volatile memory. The total number of metablocks in the storage system. It should be appreciated that a write buffer block can contain multiple metablocks. To calculate the number of sectors allocated to a write buffer block, for example, the number of metablocks containing the write buffer block can be defined as M = Ml.

Ml* M2 Metablock Size -M2 (1·0) where {N -X)* (Total.Nunber.of .Metablocks dnSystem) Metablock Size

Ml = RoundUp ^o.Nearest.Integer N * Tgc N*TO-Tgc M2 = RoundUp .to.Nearest .Integer 122731.doc -29- 1343522

Tgc is the time to perform a complete waste project collection operation. The total number of sections allocated to the write buffer block can be defined as the total number of sections = M * Metablock_Size where Μ is defined in the equation κο and MeUbl〇ck_Size is calculated for each metablock by section Size, such as 512 segments/meta block, 1〇24 segment/meta block, 2048 segment/meta block, or other size. In general, Equations 1.0 and 1.2 define that the total number of sections allocated to the write buffer block is greater than the total number of metablocks in the non-volatile memory storage system. In addition to the total number of metablocks, Equation 1 .〇 configuration—or multiple metablocks (or segments) are used to maintain additional overhead associated with writing buffer blocks (hereinafter "addition Metablock " or "Additional Section For example, the part of Equation 1〇 that defines the number of additional metablocks is (M1*M2)/(Metablock_size-M2), where the additional section can be The additional block is multiplied by the metablock size to calculate the segment configuration.

122731.doc A timeout error can be avoided by placing a number of buffer blocks as defined in Equation 1.2, even in the worst case scenario. The nest 7 -30- 1343522 is marked as invalid. In addition, the write buffer block is compressed when the write buffer block is filled or full to receive a new sector but a write buffer block clear operation may not be performed. Therefore, in a worst case scenario, when the discarding item collection time period is greater than the timeout period, the write buffer block is filled faster than the vacated rate. However, if multiple segments in the write buffer block belong to the same logical group, then all segments can be merged during an abandoned project collection operation. By arranging enough metablocks to a write buffer block, there are more segments (or entries) in the write buffer block than the total number of metablocks in the non-volatile memory storage system. And, therefore, there is a logical group with multiple active segments written into the write buffer block. When the sectors written to the buffer block are collected or merged by the obsolete item, all of the sectors belonging to the same logical group are merged into a new update block and marked as invalid in the write buffer block. Therefore, the rate at which the buffer block is vacated will be greater than the fill rate, and thereby preventing the write buffer block from being full or filled. Thus, for each given Tgc, the non-volatile memory storage system allocates a segment greater than the total number of metablocks in the non-volatile memory storage system to the write buffer block so as to prevent overshoot Time errors occur even in the worst case scenario as described above. Figure 12 is a flow diagram of the operation for temporarily storing new data in a still-scratch pad block, in accordance with an embodiment of the present invention. It should be appreciated that a non-volatile memory storage system can have multiple levels of buffering or multiple buffering levels similar to the concept of multi-level cache. In this embodiment, the write buffer block can be associated with one of a number of buffer stages. In the multi-buffer level 12273 ] .doc 31 1343522 The new data is written directly to the write buffer block, but in the other-buffer level. New capital = temporary storage of new data temporarily stored in a buffer level other than the buffer level of the write buffer block can be reduced associated with the use of the write buffer block in the staged waste collection operation t The complex (four), additional consumption 'and latent time. A short broadcast ^ ° instance is a high-speed temporary storage block. The =temporary block can be associated with a buffer level other than the write buffer block. For example, a scratch pad block can be used as a first buffer stage and a "buffer block can be used as a second buffer stage. As described above, in a single zone, in a clear case, from multiple New data for the segment of the write command can be written to different segments of the same page in the write buffer block. Because the non-volatile memory storage system may not be able to program a partial page, according to the present invention In one embodiment, new data may be temporarily stored in a first buffer level (such as a high-speed temporary storage block) before the new data is transferred to the write buffer block. The block is in the form of a data update block in which a logical segment within a logical group can be repeatedly updated in a random order and with any amount. The cache block is written by a write command. Established, where the expected logical segment does not end at the page or traverses the boundary of the physical page. The scratchpad block may contain data on the value of part of the physical page, but does not allow partial stylized page data. The scratch pad block can be maintained. _ valid page information is used Each update block in the volatile memory storage system. The non-volatile memory storage system can have, for example, eight configured update blocks, and thus, the scratch pad block can store 9 valid pages. 122731.doc -32- As shown in Figure 12, new data received from a write command is first written to a scratchpad block in operation 11〇2. The block may include multiple pages and each page includes a plurality of segments. The new data is copied to one of the γ-speed temporary storage blocks. In operation 11〇4, when the high-speed temporary storage area is in the ghost When the page associated with the write buffer block is full, new data is then copied from the scratch pad block to the write buffer block. The non-volatile memory storage system is therefore in the scratch pad block. Accumulate a page of pages & 'until all sections of the page are filled with new data from a variety of writer commands. The non-volatile memory storage system then completes the pages in a stylized operation (eg 'eight New data for the section) Copy from the high-speed temporary storage block to the write slow Because the write buffer block can be larger than the high-speed temporary storage block and can contain more valid data, the high-speed temporary storage block is less data to be collected by the abandoned project. The discarded items in the buffer block are quickly collected. Therefore, the collection of discarded items is faster than writing the new data by writing the new data to the write buffer block by means of the internal speed temporary storage block. The person to the writer buffers the H block quickly. The audio/video data-access is stored in the non-volatile memory store (four) system associated with the audio/video file = the associated ## (hereinafter: , the audio/video data ") host may need to write the video/video data at a speed of one box, the speed of the preview (compared with other data). When the host will listen to audio/video data When streaming to a non-volatile memory storage system and/or streaming audio/video data from a non-volatile memory storage system, the configuration of the ground, IL bandwidth matches or exceeds a predetermined rate. Access to audio/video data One of the obsolete items collected during the payment period 122731.doc * 33 - 1343522 Operation can degrade the write performance of audio/video data. Thus, in one embodiment, the staged item collection is performed when the multi-session write command is not associated with the audio/video material or the multi-session write command is at the beginning of an audio/video write. For & audio/video data and other materials, in one embodiment, the non-volatile memory storage system can reference the target logical address associated with the multi-session write command. Since the audio/video data is continuously written, the target logical address that is translated as a backward jump can indicate that the new data is not audio/video data (or at the beginning of the audio/video data). In another embodiment, the non-volatile memory storage system can also distinguish between audio/visual material and other data by reference to the number of segments associated with the new material. The audio/video data can be stored in a unit called a recording unit. The minimum recording unit length associated with the audio/video material can be 32 segments. The number of segments associated with the new material is not an integer multiple of 32 segments (hence) indicating that the new data is not audio/video material. The received new information that is not aligned with a recording unit or that does not begin at the beginning of a recording unit may also indicate that the new data is not audio/video material. Therefore, if one of the following conditions applies, the staged obsolete item collection can be performed: (1) the multi-session write instruction invokes the discarded item collection operation; the target logical address is translated into a backward jump; (3) the target logic The address is not aligned with a record unit boundary; and (4) after receiving the stop command (ie, the end of the multi-session write command), if the number of segments associated with the new data is not an integer multiple of 32. Single Segment Write 122731.doc • 34- U43522 Figure 13 is a flow diagram of the operation of a non-volatile memory storage system associated with a single segment write command in accordance with one embodiment of the present invention. The new data can be written to a random address on a memory cell array as a single segment. Depending on host activity and card fragments, the host can write a long file with multiple segments to a random location using a single-segment write command. Because there are a limited number of updated blocks that are configured, such single-segment writes can quickly implement the updated blocks, thereby facilitating non-volatile memory storage.

The obsolete item collection operation is performed to clear the block for subsequent write operations.

The person receives a single-segment write command in operation 12〇2 as shown in FIG. Thereafter, it is determined in operation 1204 whether the single-segment write instruction invokes an obsolete item collection operation. If the single-segment write command does not invoke a discard item collection operation, then new data is written to the memory in operation 1220. However, the right single-segment write command invokes an abandoned project collection operation', and in the operation, it is determined whether there is a pending phased project. (: Also: one has started but cannot be completed due to The collection of the project and the collection of the abandoned project in the stage - the production of the right-handed phased project collection, then continue to phase out the project in operation (10), continue the remainder of the previous abandoned project collection operation = 1210 As shown, the knife is operated as “phase-discarded project collection operation until the obsolescence project collection time period (for example, the time difference: the difference) expires, or the flight time is between. In order to fail to complete the phased project 12273I.doc -35-1343522 collection operation during the collection period of the abandoned project, the new data is written to the write in the operation 1218 == the project collection time period has not expired or can be discarded Item collection; one-item: collection operation, then in the early & slave write instruction is still called discard (four) operation 'even after completing the phased discard The same is true after the collection operation. If the single-segment write instruction does not call an obsolete item collection operation, the new data is written to the memory in = (4). On the other hand, if the single-segment write instruction is actually called If the waste project is collected, the operation collection operation is performed in operation 1214 until the waste project collection time period shown in operation 1216 is expired. If the waste project collection operation is performed during the waste project collection time period, The new data is then written to the write buffer block in operation 1218. Otherwise, the new data is written = to the sibling memory in operation 1220. Please note that if there is a staged obsolete item collection and the current write instruction In addition, an abandoned project collection operation is invoked, in operation 1216: The discarded project collection time period shown is one of the collection time periods of the abandoned project shown in operation 1210. Therefore, if there is a phased abandoned project collection and the current single zone Segment Writer Instruction Call—Abandonment project collection operation, both operations are completed within the total abandoned project collection time period. The execution time for configuring the staged obsolete item collection operation shown in operation 1208 and the deprecated item collection operation shown in operation 1214 is, for example, the difference between the timeout period and the stylized time. In operation 1220 After the new data is written to the memory, it is determined in operation 1224 whether there is a valid item and a timeout period in the write buffer block. If there is no valid item or timeout period in the write buffer block 122731.doc • 36- 'The operation ends, if there is a valid item and the timeout period ^1 period operation 1226 is executed - the writer buffer ϋ 清除 block clear operation until the timeout period expires. Also here A write buffer block compression operation is performed during the period until the timeout period expires. Multi-Segment Write Figure 14 is a flow diagram of the operation of a non-volatile memory storage system associated with a multi-session write command in accordance with an embodiment of the present invention. Write to Memory Most of the new data in the Early Hearts column is occupied by an adjacent continuous logical address space. Depending on host activity, the host can write large data archives using multisession write instructions. The new data contains a new data H with multiple segments of multiple timeout periods, an example of a multi-session writer command. In general, because multiple timeout periods are available, the new data can be written to a configured update block without being written to the write buffer block. Therefore, write buffer blocks are typically not used in multiple write instructions because collection of obsolete items can typically be done in multiple timeout periods configured for multi-segment write instructions. As shown in Figure 14, a multi-session write command is received in operation 13〇2. Thereafter, in operation 1304, new information for the segments is stored (or buffered) and the obsolete item collection operation is performed as needed. Here, the non-volatile memory storage system may not utilize a write buffer block because multiple timeout periods are configured for multi-session write instructions and an obsolete item collection operation is typically available for multi-segment write instructions. Completed during the end. Therefore, the non-volatile simon memory storage system can store new data in the RAM associated with the non-volatile memory storage system, or in the non-volatile memory storage system 122731.doc • 37· 1343522 It is associated with the reduction of the dragon towel, but it is not expected to write the person to slow down the block, and at the same time, in the section of the material between the materials, the implementation of the waste collection process. After buffering the host segment, in operation 13()6, a new data is received as to whether or not to receive less than N segments, and lUILf is again 丨. As mentioned above, it will not be more than one

The new data associated with the segment writer instruction is written to the write buffer block, because the obsolete item collection operation can usually be completed in multiple timeout periods configured for the 0 segment writer operation, in the embodiment If there is a configuration for completing the obsolete item collection operation for a sufficient timeout (four), the write buffer can be used in a multi-session write instruction to ghost. Therefore, whether or not to invoke an obsolete item collection operation '-a multi-zone & write center 7 having new data of at least N sections can be directly written by the person to update the block and not written to the write buffer block, Where N is defined as follows.

N=round [(TgC+T0)/T0] down to the nearest integer (2 〇) Equation 2_〇 shows when the write instruction is a single-segment write instruction or when non-volatile memory is stored “ When receiving new data smaller than N segments in the multi-session write command, 'write new data to the write buffer block. Referring again to Figure 14, if less than N segments are received, non-volatile The memory storage system operates according to the single-segment write command shown in FIG. 13: "," and "right receives new data of more than one segment, then = operation mo is about whether executable-write Buffer block clear operation: It is correct. For example, if the new data received in operation 13〇2 is not § abandoned project collection, or if the new data call is collected before the expiration of the configured waste project collection time period, the executable 122731 may be executed. Doc •38· Into buffer block clear operation. However, if it is bounded, it will not perform the write buffering and crying. If not! For the new data, the audio/negative lever can not be written to the right buffer to write the new data to the memory two division: if it is in the operation of the block clear operation, then in the operation (5) the write buffer area operation and - Writer Buffer $F 卩 - Staged Discarded Project Collection Execution period is performed by indecent assault until the multi-session write command is Tgc-T 总 The total busy period maintained by the memory storage system is non-volatile ttprog. Thereafter, new data is written to (for example, a "d array" in operation 1314. In this embodiment, it is possible to perform a plurality of cycles = ^Tge such as .g, four timeout periods, five timeouts: Other cycles) t perform the write buffer block shown in operation U 12 between several 13 2 before the multi-segment write command is received (eg, between 'the second sector and the third sector') After the execution of the command, in operation 1316, a determination is made as to whether the new data is an integer multiple of 32 segments. If the new data is an integer multiple of 32 segments, the new data may be For audio/video related data and the operation ends. On the other hand, if the new data is not an integer multiple of 32 segments, then in operation 1318, whether there is a valid term and a timeout period in the write buffer block Whether a determination has been made to expire. If there is no valid item in the write buffer block or the timeout period has expired, the operation ends. On the other hand, if there is a valid item and the timeout period has not expired, then Perform a write buffer block clear operation in operation 1 320 until timeout The period described above provides the method and/or system for the phased collection of abandoned projects. A waste project collection operation can be split into multiple stages 122731.doc • 39- and in multiple Multiple stages are executed on the timeout period. By splitting the discarded items, the operations are completed. Each stage of the collection process can be completed within the timeout period and thereby prevent overtime errors. The embodiments described above are described in some detail and are not limited to the details provided. There are many lanes that implement the embodiments. Therefore, the disclosed embodiments should be considered to be It is not intended to be limiting, and the embodiments are not limited to the scope of the accompanying claims, and may be modified in the scope of the appended claims.规^, otherwise the patent and the operation of the month do not imply any particular sequence of operations. [Schematic Description] FIG. 1 is a diagram of a non-volatile memory system according to an embodiment of the present invention. Simplified way of example Figure 2 is a simplified block diagram of a memory cell array organized into a plane. Circle 3 is a simplified block diagram of a memory cell page. Figure 4 is a simplified block diagram of a section of a memory cell. A simplified block diagram of a non-volatile memory storage system. A bait interface diagram is a flow diagram of a general overview of operations for phased-dump project collection in accordance with an embodiment of the present invention. Figure 7 shows a flow chart in accordance with the present invention. A simplified block diagram of an item collection operation for splitting into multiple parts of an embodiment. Figure 8 is a detailed view of the collection of items for execution - staged disposal 122731.doc -40 - 1343522 in accordance with an embodiment of the present invention. Flowchart of Operation. Figures 9A and 9B are simplified block diagrams of memory blocks having successively updated blocks collected by staged obsolescence items, in accordance with an embodiment of the present invention. 10A, FIG. 10B, FIG. 1〇c, FIG. 1D and FIG. 1 are simplified block diagrams of memory blocks having disordered update blocks collected by staged obsolescence items according to an embodiment of the present invention. . . Figure 11 is a flow diagram of an operation for optimizing access to a write buffer block in accordance with an embodiment of the present invention. Figure 2-2 is a flow diagram of an operation for temporarily storing new data in a scratchpad block in accordance with an embodiment of the present invention. FIG. 13 is a flow chart showing the operation of the non-volatile memory storage system associated with the 北 北 北 北 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 14 is a flow diagram of the operation of a non-volatile memory storage system associated with a multi-session write finger in accordance with an embodiment of the present invention. 7目关 [Main component symbol description] 102 Non-volatile memory storage system 104 Host interface 106 Processor 108 Memory interface 110 Memory controller 112 Random access memory (RAM) 114 Error correction code (ECC) circuit 116 Read Only Memory (ROM) 118 Memory 122731.doc •41 · 1343522

120 Array Logic 122 Non-Volatile Memory Cell Array 123 Memory Cell Array 124 Bus Bar 126 System Bus 202 Plane/Logic Plane 203 Plane/Logic Plane 204 Plane/Logic Plane 205 Plane/Logic Plane 210 Block 211 Block 212 Block 213 Block 220 Block 221 Block 222 Block 223 Block 401 Page 402 Section 404 Section 405 Additional Data 406 Data 502 Memory Cell Array 504 Block 122731.doc -42- 1343522

512 Logical Address Space 702 Busy Recording Signal 704 Multi-Segment Write Instruction 706 Stop Instruction 750 Timeout Period / First Timeout Period 751 Timeout Period / Second Timeout Period 752 Timeout Period / Third Time Period Period 760 New information 761 New information 762 New information 780 Part of the collection of abandoned projects / Part 1 781 Part of the collection of abandoned projects / Part II 901 New information 902 Original block A 903 New information 904 Unordered update block A 905 New data 906 New block A 907 New data 908 Write buffer block 909 New data 950 First waste project collection time period 951 Second waste project collection time period 952 Fourth waste project collection time period 122731.doc -43 - 1343522 953 Third Obsolete project collection time period / fifth abandoned project collection time period

960 original block B

962 New Update Block B/Out of Order Update Block B

964 new block B

965 Update Block C 970 First Obsolete Project Collection Time Period 972 Second Waste Project Collection Time Period

980 Original Block A 981 Existing Information

982 Continuous update block A 983 New information 984 New information

985 new original block A

986 update block C

P0 page P1 page P2 page P3 page P4 page P5 page P6 page P7 page 122731.doc -44-

Claims (1)

1343522 Patent application No. 096126798. I. Replacement of Chinese patent application scope (99 years u month) Second, ... Γ X. Application for patent garden: 1. One for operating one with a non- &amp; The system side, the Guangsheng memory cell array is non-invasive, and includes: receiving a write command to write a write command in the write-to-infrared memory cell array - Execution:, 曰 to configure a timeout period to complete the execution of a 4 &amp; project collection operation in the timeout period. Qiu Ba. Right the abandoned project collection operation helmet. n, the execution of the current waste in the timeout period; the completion of the time period, then in the rest of the Yu, the project collection operation; and after the 乂疋^, and then store the plurality of data in the A volatile memory cell array is present. A few of them are listed in the buffer. 2. As in the method of claim 1, where you are *block. , 铤 益 benefits - write buffer area 3. Kind of operation for a non-volatile ping ping open quotation. The method for recalling a body early memory memory storage system, comprising: receiving a write command to enter a plurality of data in the non-volatile, "...the hidden body is early...the write is only π, the command is configured to be a cold η Main-field u ^ λ , t ^ set the timeout period to complete one of the intrusion deductions; maintain a busy recording signal; in the -discard collection time period, part of the plurality of valid data from one or more Copying the first block to a second block; writing the plurality of data to a third block, and the second block passing through the group I to span a plurality of logical addresses, and The busy recording signal is released before the expiration of the timeout period, which causes the J22731-991I25.doc 1343522 ~~^79Sr Ήτ-^ii--------block, the second block, and the memory Body storage system _. An A block is in the non-volatile memory 4. The method of claim 3, + the non-volatile memory system t. 3 holds 4 the second block at 5, as in claim 3 The method, the middle-of-poor, the tanning part contains: a system of the plurality of valid materials The plurality of block-blocks are copied to, the part of the first plant is from the time of the - or more douche to the 5th block; and the ί::: Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ Γ The portion of the valid data is as follows: 6. The method of claim 3, further comprising: · copying the plurality of valid materials in the copy... the ... ... hai hai and the plurality of qi Before the block, it is estimated that the execution time of the execution command for the write command is deducted by 7. When the execution time exceeds the timeout period, the portion of the plurality of effect materials is copied and the plurality of blocks are copied. The data is written to the first block. The method of the person, the item 3, wherein the non-volatile memory storage system package 3 has a plurality of buffer stages, and the third block is associated with the plurality of buffer levels. The method of claim 3, wherein the writing the plurality of data comprises: writing the plurality of data to a high speed temporary storage block; and copying the plurality of data from the high speed temporary storage block to the third area Block. 122731-991125.doc 9. j % method of item 8, which Copying the plurality of data to the section associated with the 5-speed temporary storage block. 1 0 · Shikou + main book. The method of the moon clothing item 8, wherein the one and the high speed temporary storage block When the associated page is full, 'copy the plurality of data from the cache pad to the 筮r re· a ^ 乐~~ £block' page configured to include the plurality of data. The method of claim 8, wherein the cache block is associated with a first buffer, and the first block is associated with a second buffer level. Writing the plurality of data includes: writing a plurality of data to a first buffer level; and copying the plurality of data from the first buffer level to the third block, the second block of D Hai Associated with a second buffer level. The method of claim 3, wherein the discarded item collection time period is a difference between the timeout period and a stylized time associated with the writing of the plurality of data. 14. The method of claim 3, wherein the third block is a write buffer block. A method of claim 3, wherein the third block is configured to span an entire logical address space. 16. The method of claim 3, wherein the third block comprises a plurality of pages, and wherein each of the plurality of pages is indexed. The method of claim 3, wherein the third block comprises a plurality of segments, and each of the "middle" segments is indexed. 18. The method of claim 3, The write command is a single-segment write command. 122731-991125.doc mu:——. 9. for operating a non-volatile fx H η memory storage having a non-volatile memory cell array The method of the system includes: receiving a single-segment write command to write a first-plural data in the non-volatile memory unit array, the first-segment write command being configured---- a time period to complete the first execution of one of the first single-segment write commands; maintaining a first busy-recording signal; and one of the plurality of valid data from the one or more during an obsolete item collection time period Copying the first block to a second block; writing 4 the first plurality of data to a write buffer block, the write buffer block being configured to span a plurality of logical addresses; Release the first busy recording signal before the first timeout period expires; receive a second single segment write The instruction is to write the second plurality of data, and the second single-segment write command is received after the receiving the first single-segment write command, and the second single-segment write command is configured a second timeout period to complete a second execution of the second single-segment write command; maintaining a second busy-recording signal; the second portion of the plurality of valid data in the discarded item collection time period Copying the first block to the second block; and releasing the second busy record signal before the second timeout period expires] wherein the first block and the second block are In the non-volatile memory storage system. 20. The method of claim 19, further comprising the second plurality of data 122731-991125.doc -4- 21. 21. 22. 23. 24. 25. 26 27. 28. 29. Write to the write eight buffer block. The method of claim 19, further comprising writing the second plurality of data to an update block. It further includes releasing the second portion of the plurality of valid materials after the copying Or a plurality of first blocks. The method of claim 22, wherein if the second portion of the plurality of valid data is the last portion of the plurality of valid data, the one or the first region is erased The method of claim 19, further comprising copying the first plurality of data from the write eight buffer block to an update block. The method of claim 19, wherein the first plurality of data Writing to the write buffer block includes: writing the first plurality of data to a region of a fascinating temporary block &amp; 'one of the pages associated with the segment containing a plurality of regions a segment; and when the plurality of segments associated with the page are filled up, the page is resumed] to the write buffer block. The method of claim 19, wherein the write buffer block configures β for a sector level index, such as method 1 of claim 19, wherein the write eight buffer block is for a page boundary configuration. The method of claim 19, wherein the write buffer block comprises a plurality of segments and a plurality of additional segments, the plurality of segments being greater than one of the metablocks in the non-volatile suffix storage system total. A non-volatile memory storage system comprising: 122731-991125.doc 1343522 - a memory 'configured to store a storage system _· a volatile memory cell array that is buffered by a processor, And the memory and the non-volatile memory cell array are connected, the money processor is configured to execute the memory in the memory of the storage system - the following operations: receiving the intrusion into the temple order to be in the non- Volatile memory unit: The write command is configured to execute the timeout period to complete one of the 4 write commands. During the timeout period, the boom-to-segment If the waste collection operation cannot be performed within the timeout period: the waste collection operation is performed in the after-timeout period: the remainder stores the plurality of materials in the non-scheduled period before and after the timeout period In the buffer of the array of volatile memory cells. 30. A non-volatile memory storage system comprising: «remembered matter' in its K state for storing a storage system firmware; 々-non-flushedness 己 己 体 单元 , , , , , , , Writing to the buffer block, the write buffer block is configured to span the plurality of addresses; and a processor is coupled to the memory and the non-volatile memory unit array Configured to execute the 122731-991125.doc -6 - year ^ 贞 stored in the memory! =^送$赠, the storage L body contains a process of receiving-writing instructions for writing a plurality of data in the non-volatile memory cell array, the deaf button = 玄写曰令曰The configuration-timeout period is completed to complete the execution of the δ meta-write instruction, maintain a busy recording signal, and operate one part of the operation in a waste project collection time period, and the waste project collection writes the plurality of data to the write buffer. The block, and (4) release the busy signal before the expiration of the timeout period. The non-volatile memory storage system of claim 30, wherein the program instructions for use in the collection period of the waste project, the μ τ 订 亥 废弃 废弃 废弃 废弃 包含 包含 包含 包含 包含 包含彳I AM® t, at the D Hai waste project collection time w 4 T, copying a plurality of valid data to a #刀刀—or a plurality of first block stars to the first S block - A /, A child and a block are in the volatile memory storage system. 32. The non-dual (four) step-by-step of claim 30 includes a program for storing, wherein the storage is used for program instructions of the following operations: · in the execution of the waste data, writing to the write buffer The portion and the benefit block of the plurality of instructions are preceded by 'estimating an execution time of the stub for the write button 7, wherein if the execution time is to collect the portion of the operation, and the time period; Then execute the discarding buffer block. Write to the write 122731-991125.doc 1343522 33. 34. 35. 36. 37. If the non-volatile plural data of claim 30 is written to the process for the following operations, the write is slow Each of the memory storage systems in the page includes the program instructions for writing the write buffer block: writing the plurality of data to an ancient main buffer temporary block And copying the plurality of data from the high-speed temporary ir u 疋 a 八 八 &amp; block to the write buffer block. In the non-volatile memory storage system of claim 3, the buffer block includes a plurality of pages, and wherein the plurality of indices are indexed. A non-volatile memory-storage system as claimed, wherein the writer buffer block comprises a plurality of segments, and wherein each of the plurality of segments is indexed. A non-volatile memory storage system as in claim 30, wherein the write buffer block comprises a plurality of blocks. A non-volatile memory storage system comprising: a read-only memory (ROM) configured to store a storage system object; a memory cell array configured to maintain a write buffer a block, the write buffer block configured to span a plurality of logical addresses; a non-volatile memory cell array; and a processor coupled to the ROM, the memory cell array, and the non-volatile 6 The processor is configured to execute the storage system firmware stored in the ROM, the storage system firmware including program instructions for: 123731-991125.doc 1343522 receives a first write 、, · ^^ w 3 7 to write the first plurality of data to the first, 佘n*, 兀 兀 array, the first write command is configured - the first timeout The cycle is completed by the end of the dimension... the brother-write command-the first-execution; the first busy recording signal, the first one of the valid data in the collection period of the discarded project. The P-knife is clamped from one or more coin blocks to a second block, and the first plurality of rich margins are written to the write buffer block, and the first time-out is adjusted as s, w months Release the first busy recording signal before expiration. Receiving a second write phase 9 7 to write a second plurality of data to the non-volatile memory unit, the first write command being received after receiving the first write command, The second write command is set to a timeout period to complete one of the second write commands, and a second busy record signal is maintained, and the plurality of valid data are collected in the discarded item collection time period. Copying the second portion from the one or more first blocks to the second block, and releasing the second busy record signal before the second timeout period to #月, wherein the first block And the second block is in the non-volatile; memory storage system. 38. The non-volatile memory storage system of claim 37, wherein the storage system firmware further comprises program instructions for writing the second plurality of data to the write buffer block. </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> </ RTI> <RTIgt; A plurality of data is written to the program instruction of the update block. &amp; The non-volatile memory storage system of claim 37, wherein the storage system includes a program instruction for: The second part of the plurality of valid data is a trowel horse. The last part of the plurality of valid data is mourned to erase the one or more first blocks. 41. Non-volatile memory as claimed in claim 37 The storage system, wherein the storage system is further provided with program instructions for copying the first plurality of data from the write buffer block to an update block for the skeleton #β, &gt; 42. The non-volatile memory storage system of claim 37, wherein the write buffer block comprises a plurality of pages, and wherein the __th plurality of data is written to - associated with the write buffer The first stomach of the block, and the second plurality of materials are And a second page associated with the write buffer block. The non-volatile memory storage system of claim 37, wherein the write buffer block comprises a plurality of segments, and wherein The first plurality of data is written to a first sector associated with the write buffer block, and the second plurality of data is written to an associated one of the write buffer blocks The non-volatile memory storage system of claim 37, wherein the non-volatile memory system comprises a plurality of buffer stages, the write buffer block and one of the plurality of buffer stages 45. In the non-volatile memory storage system of claim 37, the first plurality of data for the 122731-991125.doc 1343522 y, · -Τ· is written to the write buffer The program instructions for the block include a command for the following operations: writing the first plurality of data to a section of the _vostatable block, where - associated with the block The page contains a plurality of sections and when the plurality of sections associated with the page are associated When the segment is filled, the pad is written to the write buffer block. 46. The non-volatile memory storage system of claim 45, wherein the high speed temporary The memory block is associated with a first buffer stage and the write buffer block is associated with a second buffer stage. a system, wherein the program instructions for writing a plurality of data for storing the plurality of data in a non-volatile memory of the request 37 include: instructing: writing the plurality of data to a first buffer stage; And copying the plurality of data from the first buffer to the write buffer block, the write buffer block being associated with the buffer level of the 坌_思思弟. 48. A non-volatile memory storage system comprising: a s-resonant body configured to save a storage system firmware with a left-handed ▲ such as / 乂; a non-volatile memory unit bundle 7C early An array configured to hold a write buffer block, the write buffer crying 6 blocks are configured to span a plurality of logical addresses; and 7 processors 'with the memory And the non-volatile memory cell array Ha. The meta-processing state is configured to execute the storage system (4) stored in the memory, the storage system 33 body containing program instructions for: 122731 -991125.doc • II - ^43522 receiving-writing instructions Writing a plurality of data in the non-volatile memory cell array, the non-volatile memory single-timeout period is performed to complete one of the write instructions, "configuring to maintain a busy recording signal, collecting time in an abandoned project Copying - part of the valid data from - or a plurality of __ blocks to the _ second block, writing the plurality of data to the write buffer block, and in the timeout period The busy signal is released before expiration, wherein the coughing first block and the second block are in the non-volatile memory storage system. 49. The non-volatile memory storage system of claim 48, wherein The program instructions for the portion of the plurality of valid data include: a program instruction for: the following operation; and tracking - for copying the portion of the plurality of valid data from the first block to the first Second block Time; and/or - stop the copying of the portion of the plurality of valid data until the time exceeds the collection time period of the discarded item. Λ 5〇 · The non-volatile memory storage system of claim 48, wherein the storage system is tough The body further includes a program instruction for: "copying the plurality of valid data and writing the data to the end" estimating an execution time of the execution of the write instruction, ^if If the execution time exceeds the material period, the portion of the plurality of materials is copied 'and the plurality of buffer blocks are used. The non-volatile memory storage system of claim 48, wherein the program instructions for writing to the portion of the plurality of tributaries are included for use in the program of claim 48. A program instruction for: providing a one-page cache pad, the page including a plurality of segments; writing the plurality of data to a section associated with the page; and when the page is full The page is copied from the scratch pad block to the write I buffer block. [52] The non-volatile memory storage system of claim 48, wherein the discarded item collection time period is one between the timeout period and a stylized time associated with the writing of the plurality of data. Difference. 53. The non-volatile memory storage system of claim 48, wherein the write buffer block is configured to span an entire logical address space. 54. The non-volatile memory storage system of claim 48, wherein the write buffer block is configured for a sector level index. The non-volatile memory storage system of claim 48, wherein the write buffer block is configured for a page boundary index. 56. The non-volatile memory storage system of claim 48, wherein the storage system firmware further comprises a program instruction for configuring a plurality of segments to the write buffer block. The segments are larger than the total number of metablocks in the non-volatile memory storage system. 122731-991125.doc -13-